Skip to content

Shehata Dataset Data Sources

Date: 2025-10-31 (Updated: 2025-11-06) Status: ✅ Complete - Data source verified and pipeline operational

Raw Data Files

The Shehata dataset preprocessing requires the following Excel files from the original paper's supplementary materials:

Required Files

File Description Source
shehata-mmc2.xlsx Supplementary Table S1: Antibody Sequences and Properties Shehata et al. 2019 Cell Reports
shehata-mmc3.xlsx Supplementary Table S2 (not used in current pipeline) Same source
shehata-mmc4.xlsx Supplementary Table S3 (not used in current pipeline) Same source
shehata-mmc5.xlsx Supplementary Table S4 (not used in current pipeline) Same source

Download Instructions

  1. Visit the Cell Reports journal page for the paper
  2. Navigate to "Supplementary Materials"
  3. Download mmc2.xlsx through mmc5.xlsx
  4. Rename them to shehata-mmc2.xlsx, shehata-mmc3.xlsx, etc.
  5. Place them in data/test/shehata/raw/ directory

Note: These Excel files are NOT committed to git (see .gitignore). They must be downloaded manually from the paper's supplementary materials.

Paper Reference

Shehata, L., et al. (2019). "Affinity Maturation Enhances Antibody Specificity but Compromises Conformational Stability." Cell Reports 28(13):3300-3308.e4.

DOI: https://doi.org/10.1016/j.celrep.2019.08.056

Note: This is the original dataset source. Novo Nordisk (Sakhnini et al. 2025) later used this dataset as a test set for their antibody non-specificity prediction models.

Processed Data

After downloading the raw Excel files, run the preprocessing scripts to generate the cleaned CSV files:

# Phase 1: Convert Excel to CSV
python3 preprocessing/shehata/step1_convert_excel_to_csv.py

# Phase 2: Extract antibody fragments
python3 preprocessing/shehata/step2_extract_fragments.py

This will generate: - data/test/shehata/processed/shehata.csv (Phase 1 output) - data/test/shehata/fragments/*.csv (16 fragment-specific files)